A hands-on walkthrough of converting Google Street View global coverage data into a spatially-sorted Parquet file using DuckDB. The process involves downloading 131 JSON files (~647 MB), importing them into DuckDB with geospatial extensions (H3, Spatial, Lindel), and exporting a Hilbert-curve-sorted, ZStandard-compressed Parquet file of 7.16 million points. The resulting dataset is then queried by year and visualized by region using QGIS with Esri basemaps, revealing Street View capture patterns across Europe, Asia, Australia, the Americas, and more.
Table of contents
My WorkstationInstalling PrerequisitesDownloading Emily's JSON FilesProducing ParquetData Fluency270 Impressions